A (relatively) simple guide to Democratic and Republican voters

As the election approaches, we’re reminded that demographic features matter in predicting voting patterns. But which ones matter most and how should they be combined?

I looked at the U.S. General Social Survey from 2002 to 2012, limited to people who reported voting for either Democrats or Republicans in presidential elections (excluding non-voters and third-party voters). Here’s the simplest way I can think of to describe the results.

(1) African Americans tended to vote for Democrats regardless of other demographic features. Overall, Democrats have won about 92% of the two-party vote from African Americans, without any other features being especially important in changing that baseline percentage.

For the rest of the voting population, the predictors differ a bit for those with at least some college education versus those who never completed a year of college.

(2) For people who are not African American and who have not completed a year of college, here’s how to produce a pretty good estimate of their presidential votes:

  • Give 1 point for each of the following that apply: (a) doesn’t go to church every week, (b) parents were immigrants, (c) Catholic, (d) not Christian (including people with no religious affiliation), (e) family income less than $75,000, (f) lesbian, gay, or bisexual, and (g) Latino, Asian, and other non-white race/ethnicity.
  • Cap that number at 5 (i.e., if it’s higher than 5, bring it down to 5). Multiply it by 12.
  • Add 19.

The result is the approximate percentage of the two-party vote going to Democrats. So, for example, for people who haven’t been to college and (1) are not frequent churchgoers, (2) have immigrant parents, (3) are Catholic, (4) have family incomes less than $75,000, and (5) are Latino, such folks voted for Democrats over Republicans around 79% of the time over the past several presidential elections (5 times 12 plus 19). On the other hand, for people who haven’t been to college and are richer, white, heterosexual, churchgoing Protestants with native-born parents, only around 19% voted for Democrats over Republicans (0 times 12 plus 19).

(3) For people who are not African American and who have completed at least a year of college, here’s how to produce their estimate:

  • Give 2 points if they are not Christian (including people with no religious affiliation).
  • Give an additional 1 point for each of the following that apply: (a) Latino, Asian, and other non-white race/ethnicity, (b) doesn’t go to church every week, (c) lesbian, gay, or bisexual, (d) never married, (e) graduate degree (MA, PhD, MD, JD, etc.), (f) female, and (g) Catholic.
  • Cap that number at 6 (i.e., if it’s higher than 6, bring it down to 6). Multiply it by 13.
  • Add 14.

Again, the result is the approximate percentage of the two-party vote going to Democrats. Former Congressman Barney Frank, for example, is Jewish (2 points), doesn’t go to synagogue every week (1 point), is gay (1 point), and has a graduate degree (1 point) — people with Frank’s demographic profile have voted for Democrats over Republicans around 79% of the time (5 times 13 plus 14). Congressman Paul Ryan, on the other hand, is Catholic, but otherwise doesn’t get any points — people with Ryan’s demographic profile have voted for Democrats over Republicans around 27% of the time (1 times 13 plus 14).

There are many ways to slice these pies. I’ve seen a lot of talk (maybe too much talk) recently about unmarried women, seniors, white evangelicals, and other categories that didn’t end up being particularly important in my models. (In the GSS data, I find that age effects are mostly covered by race and religion/church attendance, that with non-college-educated non-African Americans the effects of gender are mostly covered by income and the effects of marital status by income and religion/church attendance, that differences in religious fundamentalism are largely covered by education and church attendance, that the whiteness of white evangelicals isn’t more predictive than the whiteness of white Catholics or nonreligious whites, and so on.) But different samples can break in different ways.

Now, go forth and predict.

Actually, migrants from blue states make red states redder and blue states bluer

Robert Gebeloff and David Leonhardt have a recent article in The Upshot (NYT’s excellent online site for political analysis) titled “The Growing Blue-State Diaspora.” In it, they argue that increased migration from blue states to red states has been making red states bluer.

The analysis is based on actual migration data, but is based (as far as I can tell) on a mere assumption that migrants from blue states tend to be bluer than native voters in the red states they move to.

Is that assumption correct? Maybe not. For example, I was raised in Texas and lived there most of my life. But I’m one of those high-education liberals, people who are often uncomfortable living in Texas (other than in Austin). And so I left Texas and came to Washington DC, a place where I feel much more culturally and politically in sync. Could it be that this happens a lot — that liberal people are especially likely to leave red regions and settle in blue regions? And, likewise, are conservative people especially likely to leave blue regions and settle in red regions?

Something like this is behind the analyses of Bill Bishop and Robert Cushing in “The Big Sort.” The idea here is that mobile Americans have increasingly sorted themselves into like-minded communities.

Roughly, there are three possibilities:

(1) Diaspora: Migrants from blue states are more likely to be Democrats than natives of the red states they move to.

(2) Assimilation: Regardless of how they start off, migrants from blue states end up about as likely to be Republicans as those in the red states they move to.

(3) Big Sort: Migrants from blue states are actually more likely to be Republicans than others — after all, they chose to leave the blue states for redder pastures.

To check this, I turned to the U.S. General Social Survey (years 2000 to 2012). The GSS contains both the region in which the person currently lives (Region) as well as the region in which the person was raised (Reg16). I focused on two broad regions: (1) the Northeast (including New England and Middle Atlantic, roughly from Maine down through Pennsylvania and New Jersey) and (2) the South (including East South Central and West South Central, roughly from Tennessee to Texas).

These data support the Big Sort predictions. People who leave the Northeast are more likely to be Republicans than people raised in the Northeast who stay. People who leave the South are more likely to be Democrats than people raised in the South who stay. Similarly, people who migrate from other U.S. regions to the Northeast are more likely to be Democrats than natives of the Northeast. And people who migrate from other U.S. regions to the South are more likely to be Republicans than natives of the South.

In short, and somewhat weirdly, migrants from blue states make red states redder (and blue states bluer), and migrants from red states make blue states bluer (and red states redder). Just like I made Texas a little redder when I left and made DC a little bluer when I arrived. Migration patterns exaggerate rather than mute regional political differences as people seek out like-minded neighbors.

Genes, politics, and resistance to change

Researchers have been documenting links between genes and politics, showing in lots of cases that people’s political views have as much or more to do with their genes than with the kinds of raised-that-way factors emphasized by “socialization” theories of the past.

But what does that mean? Is there some kind of “school prayer gene” or “income redistribution gene” or “abortion rights gene” or “healthcare subsidy gene”? Of course not. So what is it about people’s genes that leads to varying political views?

In our new book, Rob Kurzban and I include a little speculation in the final chapter that maybe genes relate to things like intelligence, sociosexuality, and other relevant demographic factors, and that these genetically influenced factors affect varying political patterns in self-interested and culturally contextual ways. For an example of self-interest in action when it comes to intelligence: Some people test well and skate through higher education and thus do very well under “meritocratic” social rules. In times and places (like the contemporary U.S.) where meritocratic rules duke it out with rules based on group-based discrimination, the brainy folks tend to favor meritocratic rules, the rules that help people like them avoid being held back in social competitions. The less-brainy folks aren’t so eager to just let the smart people win — they’re much more likely to favor group-based rules that help people in their own groups and hold back people not in their own groups (based on ethnicity, religion, and whatever else is socially relevant in a given time and place).

We spend a lot of time in the book providing evidence for these kinds of self-interest claims using data from the U.S., but also confirming the overall patterns with worldwide data.

And so I was very excited to see a new article in Political Psychology, titled “Linking Genes and Political Orientations: Testing the Cognitive Ability as Mediator Hypothesis.” The authors look at political attitudes in a sample of Swedish male twins, searching for ways in which it might be true that genes affect cognitive abilities, and cognitive abilities affect political views in Sweden.

And, sure enough, they find that smart Swedes are more liberal on discrimination issues and more conservative on economic issues, and that this helps explain why genes relate to these political views.

The wheels fall off the paper when the authors speculate about why higher cognitive abilities might be associated with liberal discrimination views and conservative economic views in Sweden. They offer the wacky theory that it’s based on not-so-smart people being resistant to change. You see, immigrants bring change, and so not-so-smart people dislike immigrants. You see, Sweden has had lots of income redistribution for a while now, so pulling back from that would be change, and so not-so-smart Swedes are liberal on economic issues.

No, really, it’s a wacky theory. First, when Kurzban and I looked at worldwide data, we found, consistent with our U.S. data, that it is a very general phenomenon that less educated people like group-based discrimination and that poorer people like income redistribution. It’s safe to assume that it’s also a very general phenomenon that people with lower cognitive abilities tend to be both less educated and poorer. It’s therefore plausible to assume that people with lower cognitive abilities tend to be both conservative on discrimination issues and liberal on economic issues in lots and lots of places, not just Sweden (and, in fact, our U.S. data confirm this).

Second, looking at U.S. data, the “resistance to change” theory just fails repeatedly. For example, school prayer has been banned by the U.S. Supreme Court since the early 1960s. And yet, in our large sample, Christians with lower cognitive abilities are much more likely than Christians with higher cognitive abilities to say that they want the Supreme Court to undo (i.e., change) that long-standing anti-discriminatory policy. For another example, the U.S. has a long history of offering only very limited government help to the poor, and yet, in our large sample, people with lower cognitive abilities are more likely than people with higher cognitive abilities to want to change that long-standing reality by expanding government help for the poor. It’s just not the case that people with lower cognitive abilities routinely hold positions consistent with a dislike of policy change.

The authors offer their wacky theory, I think, for a simple reason — the obviously not-wacky theory has become practically unmentionable in some political science circles. The obviously not-wacky theory is self-interest. People who don’t win under meritocracy have self-interested reasons to favor group-based discrimination in favor of their own groups. People with lower incomes have self-interested reasons to favor income redistribution.

It makes people really uncomfortable to talk about self-interest in politics (for reasons we explore in the book). But it just becomes fluffy weirdness when (some) political scientists try to drain the real-life conflict out of political debates and turn these important policy contests into little more than spats over psychological styles — the Big Five, negativity bias, resistance to change, and so on. At some point we’re going to have to get real, as uncomfortable as that might be.

If being routinely liberal or conservative is a human universal, why is it true only of recent college-educated whites?

When people favor greater levels of income redistribution, we say their views are “liberal.” When people are pro-choice, we say their views are “liberal.” But does this mean that people who favor greater levels of income redistribution tend to be the same people as those who are pro-choice?

This has long been a source of tremendous confusion. It’s recently popped up in at least two places – Pew’s recent claim for the rise in consistently liberal and consistently conservative members of the public, and a new article on political ideology in Behavioral and Brain Sciences by Hibbing, Smith, and Alford.

I’ll cut to the chase: If one looks at data from the U.S. General Social Survey over the past 35 years on opinions on abortion and on government reduction of income differences, in fact the correlation is basically zero. Zilch. Nada. Nuthin. That is, there are as many people with ideologically mismatched views on these two issues as there are people with ideologically consistent views.

If one takes a deeper dive, it turns out that ideological coherence is higher recently, higher among whites, and higher among the well-educated. Put it all together and you get the numbers below, showing correlations (and the lines are 95% confidence intervals) for different demographic groups both in the earlier and later years of the sample.

In short, left-right coherence in the U.S. is a big deal only for college-educated whites, and then only really in recent years.


[Technical notes: For abortion, I standardized and averaged the GSS's main seven abortion items (ABDEFECT, etc.). For income redistribution, I standardized and averaged EQWLTH and GOVEQINC. In both cases, I used SPSS's "mean" command, which minimizes missing data. The resulting sample contained about 20,000 people. I split the sample into equal halves by year. "White" in this context means "non-Hispanic white," determined by combining the GSS's various race and ethnicity measures. The analyses are weighted combining the OVERSAMP and WTSSALL variables.]

Stop talking about how deep features of human nature produce “liberals” and “conservatives”

A lot of recent U.S. academic work on the nature of political issues involves online samples and samples of college students – in other words, mostly people who are college-educated whites. So, right, you look at those samples and you see higher levels of left-right coherence among different issue opinions (plus some libertarians).

But if you expand the view – look at past samples, look at non-college-educated samples, look at mostly non-white samples – the picture is very different. Here you’re about as likely to find off-axis people as left-right people.

What this means, of course, is that talk of how left-right clumping flows from “human nature” (or some related fundamental psychological something-or-other) uses a historically unusual and rather thin demographic slice to make claims about “humans” … in a situation where we actually already know that the generalization fails.

Stop saying that the main exception to the general liberal/conservative rule is libertarian

If one just looks at college-educated whites when it comes to abortion and income redistribution, in fact the largest group is “libertarian” – that is, people who are pro-choice but lean to the right on income redistribution. There are also plenty of “liberals” and “conservatives” as well. But there are very, very few people in the Region With No Agreed Name – people who are pro-life but on the left on income redistribution.

And so, researchers using online samples and college student samples often say, essentially: Most people skew “liberal” or “conservative” but there are a bunch of “libertarians” as well. (The Hibbing/Smith/Alford article I mentioned above does this, as does Jon Haidt’s recent work.)

However, look at African Americans in the general public and the biggest concentration lands in the Region With No Agreed Name, against abortion but in favor of income redistribution. In fact, there are about 3.5 times as many in that region as there are in the “libertarian” region. Among African Americans, there are almost as many in the nameless region as there are “liberals” and “conservatives” combined.

No, there’s nothing funny going on with views on abortion and income redistribution

Using GSS data limited to the 2000s, the abortion and income redistribution items both correlate substantially and roughly equally (.28 and .27, respectively) with whether people call themselves “liberal” vs. “conservative.” Looking at party identification, income redistribution has a nice correlation (.34) and abortion a smaller one (.20).

Hibbing/Smith/Alford have a discussion about how “economic issues are peripheral” in “political life” and about how “general attitudes towards taxes, social transfers, and big government” are “bloodless survey items” that “are not particularly meaningful.” It’s a bizarre set of comments given the kinds of well-known correlations I just mentioned.

Admit that we’re just not going to get off that easy

This stuff is complicated. Liberal-conservative coherence isn’t some natural necessity; it’s not even a routine feature of the modern public. We can’t just explain “liberals” and “conservatives” and be done with it. We’ve got to build up our explanations from the issues as they actually exist, with all their complicated, historically fluid, contingently coalitional, demographically related messiness.

Rob Kurzban and I try to make some progress in that direction in our new book (coming this Fall). In place of ideological clumping, we focus on something that seems to get lost in current academic discussions: These political issues aren’t just expressions of random perceptual biases or blind tribal affiliations; there’s a there there, a there that relates to the actual substance of the various issues. Public income redistribution helps people who are poorer and who lack private support (at the expense of richer people with richer social networks). The availability of family planning tools helps people who want to be sexually active while delaying having children (at the expense of people trying to minimize societal levels of casual sex). And some sexually-adventurous-but-child-delaying folks are, in fact, also richer people with richer social networks; and some casual-sex-minimizing-and-not-child-delaying folks are also poorer people with less private support. Perhaps the public’s complex, competing political positions have something to do with real life …

A reminder that when parties change, (some) voters change parties

Some researchers and commentators have been repeating the line that party preferences are strongly influenced by early political experiences and then relatively stable through later adult years. The problem is that it’s one of those claims that’s true except when it’s not.

For example, Gallup recently highlighted its polling data showing that white seniors have become substantially more Republican in party preferences since 2008 – despite assurances from some theorists that old people rarely switch parties. The parties have shifted in ways that have encouraged these changes among older whites. One of the shifts has involved Republican attempts to paint themselves as the party protecting Medicare (healthcare for seniors, who are overwhelmingly white) from the supposed cuts in the Affordable Care Act (healthcare for poorer people who are not yet seniors, many of whom are not white).

When parties shift to emphasize the views of a given group, often that group follows along. They do so notwithstanding theoretical claims that adult voters generally stay put.

Better-known examples from the last decade include white men who are either Southern or Catholic. Both white Southerners and non-Southern Catholics used to be solidly in FDR’s New Deal coalition. But then Democrats moved to the left on race (starting in the 1960s) and abortion (in the 1980s) while Republicans moved to the right on these issues.

It took decades for elected Democrats and Republicans to sort themselves out on these new dimensions. Sometimes that sorting happened through politicians switching parties, sometimes through politicians changing their old policy views, sometimes through primary defeats for old-timers who didn’t fit the new policy patterns, and sometimes through simple attrition.

These policy shifts, among others, led to substantial movements in certain voting groups’ party preferences throughout the late 20th century – particularly for white, male Southerners and white, male Catholics.

Some of the biggest shifts occurred among those born in the 1930s through the 1950s, many of whom were among the so-called “Reagan Democrats” in the 1980s, despite being well into their adult years. If we look at General Social Survey data for white men born from 1936 to 1957, for example, limiting the sample to those with at least high-school educations who were raised in the South or raised Catholic, we get the party identification levels in the graph below, which shows big shifts from the 1970s to the 1980s (when these men were in their early 30s to early 50s) and further big shifts from the 1980s to the 1990s (when these men were in their early 40s to early 60s).

GSS1947This isn’t longitudinal data, though. For that, we can turn to the Youth-Parent Socialization Study. This study began following a group of high-school seniors in 1965, who were born mostly in 1947. They were interviewed again in 1973, 1982, and 1997.

If we look at the white men who were either in the South or Catholic in 1965, we see patterns very similar to those in the GSS data. If anything, the late-age changes are more dramatic in this longitudinal sample.

long1947Here, the overall changes are actually quite modest from 1965 to 1973 (when the sample members were around ages 18 and 26, respectively), pick up pace from 1973 to 1982, and stay almost as robust from 1982 to 1997, well past ages where it is commonly assumed that people aren’t very active on the party-switching front.

In short: As the Republican Party adopted policies favored by white Southern and Catholic men, lots of these folks changed parties. Similarly, as the Republican Party has recently come to prioritize healthcare for white seniors over healthcare for younger and poorer folks, the mere fact of being non-young doesn’t prevent changes in loyalties among white seniors.

Much of the current round of talk about party stability has come in the context of thinking about future voting trends for Millennials and Latinos. But it just doesn’t make a lot of sense to assume that party preferences will be stable without also assuming that party platforms will continue to reflect these groups’ policy priorities. Further, groups’ policy priorities can change as they age. Psychological inertia won’t be enough to ensure future voting continuity from today’s young voters; the evolving policy choices of the parties will, as usual, play a central role.

No, Really, Same-Sex Marriage Doesn’t Affect Opposite-Sex Marriage

A new paper in Demography shows (in line with prior studies, e.g., here and here) that states that allow same-sex marriage do not, in fact, show reduced rates of opposite-sex marriage. This matters, presumably, because some opponents of marriage equality had predicted that allowing same-sex marriage would weaken the institution of marriage generally.

The thing is, though, does anyone really think that such findings will matter much to people otherwise opposed to same-sex marriage? For example, Jon Haidt’s work on intuitionism and moral dumbfounding makes the case that the expressed reasons people give for views on topics like same-sex marriage are often post-hoc rationalizations rather than the genuine motivations for opposition or support.

In reality, findings like those in the Demography article are part of an odd bit of policy theater, where one side provides an empirical argument (we oppose same-sex marriage because it will harm opposite-sex marriage) that the other side knows isn’t a real reason for their policy views, the other side goes about falsifying the claim anyway (pretending that they take the argument seriously), and the one side either ignores the finding or claims the finding is flawed or points to a competing study or simply moves on to some other post-hoc rationalization, without anyone’s policy judgments having changed.

Simple summary of “Dog-Whistle Politics: Multivocal Communication and Religious Appeals”

People who say they don’t want politicians to be explicitly religious don’t like politicians who are explicitly religious, but only if they are aware of the fact that the politician is being explicitly religious.

The paper: Albertson, B. L. (2014). Dog-whistle politics: Multivocal communication and religious appeals. Political Behavior.

Diversity on campus (sex, drugs, and tattoos edition)

A new study out in The Social Science Journal by Keith King and Rebecca Vidourek shows something that everybody already knows: Party kids are more likely than abstinent virgins to get tattoos. In a sample of 998 students from a Midwestern university (they don’t say which one), almost 30% had tattoos. Consistent with prior research (as well as Wedding Crashers), the inked kids were more likely to drink, smoke, and use drugs, and especially more likely to be sexually active than their non-inked classmates.

There are various other details, of course: Women were more likely to have tattoos than men; tattooed students used birth control more than non-tattooed kids, something that cuts against their reckless image; and so on. The study is also a good example of the folly of asking people why they do the things they do — turns out the major reasons people give for getting tattoos include super-illuminating things like it being fun.

Another point embedded in the data relates to media portrayals of college students. They’re all a bunch of promiscuous partiers, right?

Well, no. I constructed the accompanying chart from the data presented in the paper. It shows the relative size of groups from the sample, splitting them up by those who have had sex and those who are virgins, and, among those who have had sex, those who drank or used drugs the last time they had sex versus those who last had sex while stone-cold sober.


If you take the anecdotal methods of click-hungry journalism seriously, you might think that the majority of the sample would be in the partied-the-last-time-I-had-sex category and that very few would be virgins. In fact, these groups are close to the same size, with a slight lead for the virgins: 19% were virgins and 15% used alcohol or drugs the last time they had sex. The other two-thirds were in the too-boring-to-write-about middle — they had had sex before, but didn’t last have sex in the context of partying.

The important story missed by most articles about those crazy kids is the tremendous variation on campus when it comes to what the literature likes to call “risky behaviors.” In data I’ve collected over the years, I find that there’s generally about a fifth or a quarter of college kids who do extremely little by way of sex/drinking/drugs, balanced out by a similar-sized group made up of the kinds of students you’ll typically see profiled in stories — regular weekend drinkers who have had a number of hook-up partners.

So, you know, don’t believe the hype and all that. But also, keep your eye on the big theme, which is the impressive diversity of modern lifestyle patterns.

The study: King, K. A., & Vidourek, R. A. Getting inked: Tattoo and risky behavioral involvement among university students. The Social Science Journal (2013), http://dx.doi.org/10.1016/j.soscij.2013.09.009

Article on political socialization buries the lede

A recent article in Psychological Bulletin looks at various studies on the resemblance between parents and children when it comes to discriminatory political attitudes. The article is framed as investigating socialization accounts — accounts claiming that people get their varying political attitudes by being raised in different ways. The authors find that, indeed, there really is a correlation between the attitudes of parents and children, and, thus, that one of the big roadblocks to socialization theories has been cleared away: “The general argumentation that parental attitudes are irrelevant for prejudice formation in children because of an absence of a parent–child correlation can thus no longer be maintained.”

But then, tucked away at the end, is this:

[F]inally, we must address the question whether significant indices of parent–child similarities even in methodologically rigorous longitudinal studies can be interpreted as reflecting socialization effects (be it parent–child, child–parent, or environment–parent/child) given the possibility of genetic influences. Because effects of parental genes and parental attitudes and behaviors are correlated, ostensible influences of parental attitudes may actually be artifacts of genetic influences. Several twin studies provide first evidence that intergroup attitudes and related attitudes, such as social conservatism, social dominance orientation, right-wing authoritarianism, and political attitudes, have modest to substantial heritability coefficients. Notably, these heritability indices are similar in magnitude to the effect sizes we have found in the current metaanalysis. Furthermore, the shared environment of siblings—which would include parental attitudes and behaviors—appears to be of negligible importance in these studies (as compared to the unshared environment). This could be taken to imply that parents have little direct influence on their children’s intergroup attitudes, or at least that they do not affect different siblings in the same ways.

In other words: Yeah, well, we’ve been telling you for 20 pages about how important parent-child socialization is for political attitudes, but it turns out that adults’ attitudes probably don’t have much to do with how they were raised.

It has the feel of something that probably wasn’t in the authors’ submitted draft, but was added at the insistence of a reviewer/editor later in the process (I could well be wrong about that, as I have no inside information here).

Behavioral genetics findings have thrown a monkey wrench in traditional social science. We’re at that awkward stage where everyone knows it, but there’s so much inertia behind the old approaches that we’re likely to see many more weird examples like this of burying the lede.

The paper: Degner, J., & Dalege, J. (2013). The apple does not fall far from the tree, or does it? A meta-analysis of parent-child similarity in intergroup attitudes. Psychological Bulletin, 6, 1270-1304.